Continuous charging assignment algorithm for heterogeneous robot clusters based on E-CARGO

Abstract

Forming robot clusters is a common means to improve work efficiency and ensure personnel safety in intelligent logistics warehousing. However, designing a reasonable charging sequence and time for each robot in the cluster poses a major challenge for management personnel. In actual scenarios, the working and charging processes of the robots must be considered simultaneously. This study also considers multiple constraints on robot battery levels and available workstations under different parameters to ensure each team operates normally at any given time. Additionally, to enhance cooperation efficiency and alleviate the workload of management personnel, the problem is transformed from conventional one-time assignments to adaptive continuous assignments within fixed time periods. Accordingly, the ACP-GRAC problem is proposed. Firstly, a group role assignment model with conflicts is introduced to resolve assignment conflicts caused by mismatches between job skills and position requirements. Then, energy information is shared among robots during the charging process, adding adaptive cooperation concepts to address continuous charging issues. Finally, the E-CARGO model is utilized to formalize the ACP-GRAC problem, and the HCCA algorithm is designed for its solution. Furthermore, a classification backtracking strategy is proposed to simultaneously address the constraint of continuous operation at any given time and the problem of constraint violations caused by the HCCA algorithm’s pursuit of maximum team performance. Moreover, since the classification backtracking strategy can compute the theoretical minimum remaining energy of the robot before departing for charging, it further optimizes the overall work efficiency. Additionally, this study addresses the issue of low cluster work efficiency due to the unreasonable initial planning sequence of the algorithm. In the experimental section, comparisons are made with recent excellent heuristic algorithms and deep reinforcement learning algorithms. Experimental results demonstrate that B-HCCA exhibits significant advantages in both solution efficiency and solution quality when addressing adaptive cooperation problems with group role assignment conflicts.